Identifikasi Penyakit Gigi dan Mulut Menggunakan Metode Mobilenetv3 dan Efficientnetv2

Abstract

Dental and oral diseases remain a prevalent health issue in Indonesia. The lack of routine dental check-ups and limited number of medical professionals highlight the need for an automated image-based identification system. This study implements two deep learning architectures, MobileNetV3 and EfficientNetV2, to classify seven types of dental and oral diseases based on clinical images. The dataset, consisting of 13,138 images sourced from Kaggle and the Dental and Oral Hospital of Universitas Sumatera Utara, was split into training, validation, and testing sets. The Models were trained using various hyperparameter combinations and evaluated with F1-score, precision, and recall. The best result was achieved by the MobileNetV3Large architecture with a test Accuracy of 0.949. This research demonstrates the effectiveness of the proposed method for disease classification and its potential for deployment in AI-based diagnostic systems.